During Edmonds College talk, virologist explains how AI is changing gene therapy research

Dr. Semih Tareen speaks at Edmonds College Feb. 7.

If artificial intelligence can help physicians predict who is more or less likely to get a medical condition, it may also help molecular biologists predict which gene or gene mutations are more likely to cause a specific disease. During a presentation at Edmonds College last week, Dr. Semih Tareen – a virologist and former executive director of Sana Biotechnology – discussed how gene therapy and AI are changing biomedicine.

Dr. Tareen is the first guest in Edmonds College’s Community Connections Speaker Series featuring leaders in the AI and data science industry. The free speaker series will bring experts to campus to share their knowledge and innovations, allowing community members to see real-world AI and data science applications.

Scientists use gene therapy to try to treat diseases by changing the instructions found in the DNA. They do this by adding, removing or changing genes that likely increase the risk of certain people getting specific diseases, notably type I diabetes, cancer and lupus. It is like giving your cells new tools to repair themselves and stay healthy.

“We are taking different antibody sequences that recognize T cells [a type of white blood cell],” Tareen said. “We are adding them to our virus so that our viruses will infect [those] cells in the body. So we have to screen the body and make sure this virus doesn’t go anywhere else –  the brain or lungs or liver or anywhere else where it would cause a problem.”

“And most importantly, we have to make sure this virus doesn’t go to the germ cells, sperm cells or egg cells,” he emphasized.

Tareen said that if someone were to be infected with a virus and survives, and they were then to have children, their children and future descendants will carry that viral genome sequence forever. In fact, about 50% of the human genome is made up of ancient viral sequences that have been passed on from early humans who have been infected with viruses millions of years ago, he said. 

“It’s permanent,” Tareen said. “[But] we don’t want this to happen on a clinical side. We don’t want to cause this so we control everything to prevent this from happening. But in an evolutionary sense, it has already happened, so all of us are half virus for that reason.”

In one example of how gene therapy works, Tareen described how a drug called lisocabtagene maraleucel – or Liso-cel – is used to treat blood cancers.

“We isolate T cells and in the lab, we take those T cells and add the viruses that my team and I have been preparing,” he said. “We call these viruses viral vectors because viruses imply something that might make you sick or something that spreads. These don’t spread. They are a dead-end product, so you add them to the cells. They deliver the genetic information, they enter the cell, travel through the cytoplasm, enter the nucleus where they integrate that genetic information into the cell so they’re part of that patient’s genome in that one cell.”

With the viral genetic code in place, Tareen said that he and his team hope that the cell would duplicate itself during the cultivation process. When the cells reach an ideal number, they check these cells to make sure that the gene delivery worked and that the cells express that “special protein,” which identifies tumors and kills them. 

“We rely on antibody sequences that have been developed to recognize tumors over the years,” Tareen said. “We just take one of those and then engineer it into our virus and then we take other chunks of information from over the years and assemble it into a chimeric antigen receptor (CAR).” CAR-T cell therapy is used to treat B cells in leukemia and lymphoma.

Despite the progress of gene therapy and AI, government regulatory bodies and agencies are playing catch-up to the research, Tareen said.

“We then call the patient back and then we deliver these cells back into the patient,” Tareen continued. “These patients usually stay in the hospital [intensive care unit] because they could get really sick. Why? Because they end up with what’s called a cytokine storm. 

“For example, if I get the ‘man flu’ where I feel so bad I can’t even play my Xbox, that’s my body fighting against the infection by releasing cytokine, and that inflammation of cytokine release is exactly what my body has evolved to do,” he continued. “It does it well, but it makes me feel horrible. Imagine all of a sudden these patients get millions of cells that recognize tumors and kill them in the blood just like that and they release all these enzymes and that creates that cytokine storm.”

In the worst-case scenario, Tareen said that patients can get into a state of “neurotoxicity” where their brain becomes catatonic and they cannot even talk. Tareen said that the cytokine storm is also the same reason why COVID-19 is so deadly. “The virus doesn’t kill you. It’s your immunity that kills your lungs,” he said.

The AI software that Tareen used examines markers on T cells to understand if they are able to kill the tumors. Then the AI creates many models for dosing the patient as the cells are being administered via an intravenous line. 

“AI is being developed to see if the prognosis for these patients will be good, depending on the treatment outcome,” Tareen said. “The challenge is the regulatory authorities, like the FDA, to catch up, and we work with them for all of our clinical trials and our drugs to be approved. For example, you will submit a thousand or so pages of research just to be able to begin a clinical trial of 10 people. Anytime AI is involved, [it’s] going to make predictions and based on those predictions, medical staff is going to change the course of treatment. Based on the results, insurance companies decide if they will cover that treatment.”

Despite the progress gene therapy and AI have made, government regulatory bodies and agencies are constantly playing catch-up to the research. Tareen said that by the time researchers are ready to apply their work to the clinic, they have to wait for the FDA and other agencies to give them the green light.

“I’ve been working with the FDA since 2010 where I have to be on the call with them to explain to them what these therapies are, to answer their questions on safety issues,” he said. “Because cell and gene therapy is advancing so fast with so much potential, they have been able to address this back by saying that [the] cell and gene therapy approval process has to be fast, and fast enough to make sure the patient is still safe but to also make sure that patients are not held back from any potential therapy that can save their lives.”

You can watch a video of Tareen’s presentation here.

— Story and photos by Nick Ng

    1. Good question, Niko. Dr. Tareen didn’t specify *how* AI is used for gene therapy in terms of the process, but he did mention what it’s used for and abstractly described how it was done. The video at the end of the article has the entire lecture where I only highlighted what I thought was the most relevant.

      I suspect that the *how* may be a bit too technical or esoteric for the public talk series. But if the audience was almost entirely biology students and life science professors, I think he would describe the how more, just like the methodology section in research papers where most laypeople (like me) would glaze over because most of the methods are jargony and is written for other scientists, not the public.

Leave a Reply

Your email address will not be published. Required fields are marked *

Real first and last names — as well as city of residence — are required for all commenters.
This is so we can verify your identity before approving your comment.

By commenting here you agree to abide by our Code of Conduct. Please read our code at the bottom of this page before commenting.